espnet3.systems.base.system.BaseSystem

source

class espnet3.systems.base.system.BaseSystem(training_config: DictConfig | None = None, inference_config: DictConfig | None = None, metrics_config: DictConfig | None = None, publication_config: DictConfig | None = None, stage_log_mapping: dict | None = None)

Bases: object

Base class for all ESPnet3 systems.

DATASET_BUILDER_CLASS_NAME

Name of the builder class expected in each dataset module (default "DatasetBuilder").

DATASET_CLASS_NAME

Name of the dataset class expected in each dataset module (default "Dataset"). Used by subclasses that instantiate datasets directly.

Expected stage methods. : - create_dataset()

• train()
• infer()
• measure()
• publish()
• pack_model()
• upload_model()

All behavior is config-driven.

• Parameters:
  • training_config (DictConfig | None) – Training configuration.
  • inference_config (DictConfig | None) – Inference configuration.
  • metrics_config (DictConfig | None) – Measurement configuration.
  • stage_log_mapping (dict | None) – Optional overrides for stage log path resolution. Keys are stage names; values are dotted attribute paths (e.g., "training_config.exp_dir") or lists/tuples of such paths (first non-empty value wins).

Base stage log mapping. : - create_dataset -> training_config.data_dir

• collect_stats -> training_config.stats_dir
• train -> training_config.exp_dir
• infer -> inference_config.inference_dir
• measure -> metrics_config.inference_dir
• pack_model -> training_config.exp_dir
• upload_model -> training_config.exp_dir

Any stage missing from the mapping (or resolving to None) falls back to the default log directory: training_config.exp_dir when available, otherwise <cwd>/logs.

Examples

Override a subset of stage log paths.

system = BaseSystem(
    training_config=train_cfg,
    inference_config=infer_cfg,
    metrics_config=measure_cfg,
    stage_log_mapping={
        "infer": "training_config.exp_dir",
        "measure": "training_config.exp_dir",
    },
)

Initialize the system with optional stage configs.

• Parameters:
  • training_config – Training configuration for data preparation, statistics collection, and model training.
  • inference_config – Inference configuration used by the infer stage.
  • metrics_config – Measurement configuration used by the measure stage.
  • publication_config – Publication configuration for pack_model and upload_model stages.
  • stage_log_mapping – Optional per-stage log directory overrides.

DATASET_BUILDER_CLASS_NAME

*= 'DatasetBuilder'*

DATASET_CLASS_NAME

*= 'Dataset'*

collect_stats(*args, **kwargs)

Collect statistics needed for training.

create_dataset(*args, **kwargs)

Create datasets from dataset references.

infer(*args, **kwargs)

Run inference on the configured datasets.

measure(*args, **kwargs)

Compute evaluation metrics from hypothesis/reference outputs.

pack_demo(*args, **kwargs)

Pack demo artifacts into a runnable demo bundle.

pack_model(*args, **kwargs)

Pack model artifacts into an espnet3 bundle.

train(*args, **kwargs)

Train the system model.

upload_demo(*args, **kwargs)

Upload demo bundle to HuggingFace Spaces (stub).

upload_model(*args, **kwargs)

Upload model bundle to HuggingFace.